東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Adaptive Object Detection for Real-T...

Liu, Yu,

FindBook

Google Book

Amazon

博客來

Adaptive Object Detection for Real-Time Traffic Monitoring /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Adaptive Object Detection for Real-Time Traffic Monitoring // Yu Liu.
作者:	Liu, Yu,
面頁冊數:	1 electronic resource (78 pages)
附註:	Source: Dissertations Abstracts International, Volume: 86-07, Section: A.
Contained By:	Dissertations Abstracts International86-07A.
標題:	Computer science. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31762173
ISBN:	9798302180506

Adaptive Object Detection for Real-Time Traffic Monitoring /
Liu, Yu,

Adaptive Object Detection for Real-Time Traffic Monitoring /Yu Liu. - 1 electronic resource (78 pages)

Source: Dissertations Abstracts International, Volume: 86-07, Section: A.

Real-time object detection is crucial in traffic management video applications, such as intelligent transportation systems and surveillance, where deep learning models have significantly improved detection accuracy. However, as these models become more complex, achieving a frame processing speed of at least 30 frames per second (fps) for real-time performance has become increasingly challenging. To address this issue, we investigate three approaches focused on Adaptive Object Detection for Real-Time Traffic Monitoring. The first approach, ERD (Empty Road Detection), is a streamlined lightweight Convolutional Neural Network (CNN) model that preprocesses video frames to filter out those without key objects like cars, motorcycles, or pedestrians. By processing only non-empty frames, ERD increases overall frame processing speed without sacrificing detection accuracy. Nevertheless, ERD's requirement to categorize each frame in the video sequence might be redundant, given that changes in road conditions between empty and occupied typically occur gradually rather than abruptly. To improve efficiency even further, we propose a second method, L-filter, which builds on ERD by incorporating a Long Short-Term Memory (LSTM) module. L-filter uses a hybrid framework that alternates between periodic ERD operations and LSTM-driven predictions, reducing the need for intensive pixel-level analysis. This approach further reduces computational load while maintaining detection accuracy. However, both ERD and L-filter encounter challenges in scenarios like traffic congestion, where objects are stationary, leading to the unnecessary processing of frames. To overcome this, we propose a third approach that estimates object movement speed in pixels per frame. This method refines traditional optical flow techniques by employing a multilayer perceptron to accurately estimate pixel movement speed, allowing for the elimination of redundant frames. Together, these methods improve frame processing rates and scalability for real-time object detection systems without compromising accuracy, enabling a single system to monitor multiple traffic directions simultaneously, making them highly suitable for practical traffic video processing applications.

English

ISBN: 9798302180506Subjects--Topical Terms:

523869
Computer science.
Subjects--Index Terms:

Convolutional Neural Network

Adaptive Object Detection for Real-Time Traffic Monitoring /
LDR:03639nmm a22004093i 4500 001 2431675
005 20250616055554.5
006 m o d
007 cr|nu||||||||
008 261011s2024 miu||||||m |||||||eng d
020 $a 9798302180506
035 $a (MiAaPQD)AAI31762173
035 $a AAI31762173
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Liu, Yu, $e author. $3 3708607
245 1 0 $a Adaptive Object Detection for Real-Time Traffic Monitoring / $c Yu Liu.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2024
300 $a 1 electronic resource (78 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 86-07, Section: A.
500 $a Advisors: Kang, Kyoung-Don Committee members: Zhang, Zhongfei; Blackburn, Jeremy.
502 $b Ph.D. $c State University of New York at Binghamton $d 2024.
520 $a Real-time object detection is crucial in traffic management video applications, such as intelligent transportation systems and surveillance, where deep learning models have significantly improved detection accuracy. However, as these models become more complex, achieving a frame processing speed of at least 30 frames per second (fps) for real-time performance has become increasingly challenging. To address this issue, we investigate three approaches focused on Adaptive Object Detection for Real-Time Traffic Monitoring. The first approach, ERD (Empty Road Detection), is a streamlined lightweight Convolutional Neural Network (CNN) model that preprocesses video frames to filter out those without key objects like cars, motorcycles, or pedestrians. By processing only non-empty frames, ERD increases overall frame processing speed without sacrificing detection accuracy. Nevertheless, ERD's requirement to categorize each frame in the video sequence might be redundant, given that changes in road conditions between empty and occupied typically occur gradually rather than abruptly. To improve efficiency even further, we propose a second method, L-filter, which builds on ERD by incorporating a Long Short-Term Memory (LSTM) module. L-filter uses a hybrid framework that alternates between periodic ERD operations and LSTM-driven predictions, reducing the need for intensive pixel-level analysis. This approach further reduces computational load while maintaining detection accuracy. However, both ERD and L-filter encounter challenges in scenarios like traffic congestion, where objects are stationary, leading to the unnecessary processing of frames. To overcome this, we propose a third approach that estimates object movement speed in pixels per frame. This method refines traditional optical flow techniques by employing a multilayer perceptron to accurately estimate pixel movement speed, allowing for the elimination of redundant frames. Together, these methods improve frame processing rates and scalability for real-time object detection systems without compromising accuracy, enabling a single system to monitor multiple traffic directions simultaneously, making them highly suitable for practical traffic video processing applications.
546 $a English
590 $a School code: 0792
650 4 $a Computer science. $3 523869
650 4 $a Information science. $3 554358
653 $a Convolutional Neural Network
653 $a Object detection
653 $a Real-time performance
653 $a Road detection
690 $a 0984
690 $a 0723
710 2 $a State University of New York at Binghamton. $b Computer Science. $e degree granting institution. $3 3820598
720 1 $a Kang, Kyoung-Don $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 86-07A.
790 $a 0792
791 $a Ph.D.
792 $a 2024
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31762173